Past events

Retinal prostheses represent an exciting development in science, engineering, and medicine – an opportunity to exploit our knowledge of neural circuitry and function to restore or even enhance vision. However, although existing retinal prostheses demonstrate proof of principle in treating incurable blindness, they produce limited visual function. Some of the reasons for this can be understood based on the exquisitely precise and specific neural circuitry that mediates visual signaling in the retina. Consideration of this circuitry suggests that future devices may need to operate at single-cell, single-spike resolution in order to mediate naturalistic visual function. I will show large-scale multi-electrode recording and stimulation data from the primate retina indicating that, in some cases, such resolution is possible. I will also discuss cases in which it fails, and propose that we can improve artificial vision in such conditions by incorporating our knowledge of the visual system in bi-directional devices that adapt to the host neural circuity. Finally, I will briefly discuss the potential implications for other neural interfaces of the future.

In classic models of vision, vision proceeds in a hierarchical fashion, from low-level analysis (edges and lines) to figural processing (shapes and objects). Low-level processing determines high-level processing. Here, we show that shape processing determines basic visual processing as much as the other way around. For example, we presented a vernier stimulus and asked observers to indicate its offset direction. Performance strongly deteriorated when the vernier was surrounded by a square, in line with most models of vision. Surprisingly, performance improved when more squares were added. This improvement of performance can hardly be explained by classic models of vision, which predict a further deterioration of performance. We propose that shape interactions precede low-level processing in a recurrent fashion. Using high density EEG and trans-cranial magnetic stimulation (TMS), we show how good Gestalt emerges during recurrent, unconscious processing within 420ms. The outcome of this processing, i.e., the conscious percept, determines, paradoxically, what is usually referred to as early visual processing.

Retinal degenerative diseases lead to blindness due to loss of the “image capturing” photoreceptors, while neurons in the “image-processing” inner retinal layers are relatively well preserved. Information can be reintroduced into the visual system using electrical stimulation of the surviving inner retinal neurons. Some electronic retinal prosthetic systems have been already approved for clinical use, but they provide low resolution and involve very difficult implantation procedures. We developed a photovoltaic subretinal prosthesis which converts light into pulsed electric current, stimulating the nearby inner retinal neurons. Visual information is projected onto the retina from video goggles using pulsed nearinfrared (~880nm) light. This design avoids the use of bulky electronics and wiring, thereby greatly reducing the surgical complexity. Optical activation of the photovoltaic pixels allows scaling the implants to thousands of electrodes. In preclinical studies, we found that prosthetic vision with subretinal implants preserves many features of natural vision, including flicker fusion at high frequencies (>20 Hz), adaptation to static images, antagonistic center-surround organization and nonlinear summation of subunits in receptive fields, providing high spatial resolution. Results of the clinical trial with our implants (PRIMA, Pixium Vision) having 100µm pixels, as well as preclinical measurements with 75 and 55µm pixels, confirm that spatial resolution of prosthetic vision can reach the sampling density limit. For a broad acceptance of this technology by patients who lost central vision due to age-related macular degeneration, visual acuity should exceed 20/100, which requires pixels smaller than 25µm. I will describe the fundamental limitations in electro-neural interfaces and 3-dimensional configurations which should enable such a high spatial resolution. Ease of implantation of these wireless arrays, combined with high resolution opens the door to highly functional restoration of sight.

Driven by recent technological advancements, behavior and brain activity can now be measured at an unprecedented resolution and scale. This “big-data” revolution is akin to a similar revolution in biology. In biology, the wealth of data allowed systems-biologists to uncover the underlying design principles that are shared among biological systems. In my studies, I apply design principles from systems-biology to cognitive phenomena. In my talk I will demonstrate this approach in regard to creative search. Using a novel paradigm, I discovered that people’s search exhibits exploration and exploitation durations that were highly correlated along a line between quick-to-discover/quick-to-drop and slow-to-discover/slow-to-drop strategies. To explain this behavior, I focused on the property of scale invariance, which allows sensory systems to adapt to environmental signals spanning orders of magnitude. For example, bacteria search for nutrients, by responding to relative changes in nutrient concentration rather than absolute levels, via a sensory mechanism termed fold change detection (FCD). Scale invariance is prevalent in cognition, yet the specific mechanisms are mostly unknown. I found that an FCD model best describes creative search dynamics and further predicts robustness to variations in meaning perception, in agreement with behavioral data. These findings suggest FCD as a specific mechanism for scale invariant search, connecting sensory processes of cells and cognitive processes in human. I will end with a broader perspective and outline the benefits of the search for computational design principles of cognition.

I will discuss recent work on the neural circuitry underlying model-free and model-based reinforcement learning (RL). While there has been considerable focus on dopamine and its action in the striatum, particularly for model-free RL, our recent work has shown that the amygdala also plays an important role in these processes. We have further found that the amygdala and striatum learn in parallel. However, the amygdala learns more rapidly than the striatum. Therefore, each structure tends to be optimized for different reward environments. Overall, the work in our lab outlines roles for multiple neural circuits spanning cortical-basal ganglia-thalamocortical loops, as well as the amygdala’s interaction with these circuits, in RL.

Sensory input reaching the brain from bilateral and offset channels is nonetheless perceived as unified. This unity could be explained by simultaneous projections to both hemispheres, or inter-hemispheric information transfer between sensory cortical maps. Odor input, however, is not topographically organized, nor does it project bilaterally, making olfactory perceptual unity enigmatic. Here we report a circuit that interconnects mirror-symmetric isofunctional output cells between the mouse olfactory bulbs. Connected neurons respond to similar odors from ipsi- and contra-nostrils, whereas unconnected neurons do not respond to odors from the contralateral nostril. This circuit enables sharing of odor information across hemispheres in the absence of a cortical topographical organization, suggesting that olfactory glomerular maps are the equivalent of cortical sensory maps found in other senses.

Due to the immense complexity of the human brain, the study of its development, function, and dysfunction during health and disease has proven to be challenging. The advent of patient-derived human induced pluripotent stem cells, and subsequently their self-organization into three-dimensional (3D) brain organoids, which mimics the complexity of the brain's architecture and function, offers an unprecedented opportunity to model human brain development and disease in new ways. However, there is still a pressing need to develop new technologies that recapitulate the long-term developmental trajectories and the complex in vivo cellular environment of the brain. To address this need, we have developed a human brain organoid-based approach to generate a chimeric human/animal brain system that facilitates long-term ana! tomical integration, differentiation, and vascularization in vivo. We also demonstrated the development of functional neuronal networks within the brain organoid and synaptic-cross interaction between the organoid axonal projections and the host brain. This approach set the stage for investigating human brain development and mental disorders in vivo, and run therapeutic studies under physiological conditions.